T
E
L
KO
M
N
I
KA
T
e
lec
om
m
u
n
icat
ion
,
Com
p
u
t
i
n
g,
E
lec
t
r
on
ics
an
d
Cont
r
ol
Vol.
18,
No.
3,
J
une
2020,
pp.
14
3
9
~
14
4
6
I
S
S
N:
1693
-
6930,
a
c
c
r
e
dit
e
d
F
ir
s
t
G
r
a
de
by
Ke
me
nr
is
tekdikti
,
De
c
r
e
e
No:
21/E
/KP
T
/2018
DO
I
:
10.
12928/
T
E
L
KO
M
NI
KA
.
v18i3.
14381
1439
Jou
r
n
al
h
omepage
:
ht
tp:
//
jour
nal.
uad
.
ac
.
id/
index
.
php/T
E
L
K
OM
N
I
K
A
Im
age
d
e
n
osi
n
g i
n
u
n
d
e
r
w
a
t
e
r
ac
ou
s
t
ic
n
oi
s
e
u
si
n
g
d
is
c
r
e
t
e
w
ave
le
t
t
r
a
n
sf
or
m
w
it
h
d
if
f
e
r
e
n
t
n
oi
se
l
e
v
e
l
e
st
ima
t
io
n
Yas
in
Yous
if
Al
-
Aboos
i
,
Radh
i
S
e
h
e
n
I
s
s
a,
Ali
k
h
ali
d
Jas
s
im
Facu
l
t
y
o
f
E
n
g
i
n
eeri
n
g
,
U
n
i
v
e
rs
i
t
y
o
f
Mu
s
t
a
n
s
i
ri
y
ah
,
Ira
q
Ar
t
icle
I
n
f
o
AB
S
T
RA
CT
A
r
ti
c
le
h
is
tor
y
:
R
e
c
e
ived
Oc
t
23
,
2019
R
e
vis
e
d
F
e
b
29
,
2020
Ac
c
e
pted
M
a
r
12
,
2020
In
man
y
a
p
p
l
i
ca
t
i
o
n
s
,
Imag
e
d
e
-
n
o
i
s
i
n
g
an
d
i
mp
r
o
v
eme
n
t
rep
re
s
en
t
es
s
en
t
i
a
l
p
ro
ce
s
s
e
s
i
n
p
re
s
en
ce
o
f
co
l
o
red
n
o
i
s
e
s
u
ch
t
h
a
t
i
n
u
n
d
er
w
at
er.
Po
w
er
s
p
ec
t
ral
d
e
n
s
i
t
y
o
f
t
h
e
n
o
i
s
e
i
s
c
h
an
g
eab
l
e
w
i
t
h
i
n
a
d
efi
n
i
t
e
freq
u
e
n
c
y
ra
n
g
e,
an
d
au
t
o
co
rre
l
at
i
o
n
n
o
i
s
e
fu
n
c
t
i
o
n
i
s
d
o
e
s
n
o
t
l
i
k
e
d
e
l
t
a
fu
n
ct
i
o
n
.
So
,
n
o
i
s
e
i
n
u
n
d
er
w
at
er
i
s
c
h
aract
er
i
zed
a
s
co
l
o
re
d
n
o
i
s
e.
In
t
h
i
s
p
ap
er,
a
n
o
v
e
l
i
ma
g
e
de
-
n
o
i
s
i
n
g
met
h
o
d
i
s
p
ro
p
o
s
ed
u
s
i
n
g
mu
l
t
i
-
l
ev
e
l
n
o
i
s
e
p
o
w
er
es
t
i
ma
t
i
o
n
i
n
d
i
s
cret
e
w
av
e
l
et
t
ra
n
s
f
o
rm
w
i
t
h
d
i
ffere
n
t
b
a
s
i
s
fu
n
c
t
i
o
n
s
.
Peak
s
i
g
n
al
t
o
n
o
i
s
e
ra
t
i
o
(PSN
R)
an
d
mean
s
q
u
are
d
erro
r
rep
r
es
en
t
ed
p
erfo
rma
n
ce
meas
u
re
s
t
h
a
t
t
h
e
res
u
l
t
s
o
f
t
h
i
s
s
t
u
d
y
d
e
p
en
d
o
n
i
t
.
T
h
e
res
u
l
t
s
o
f
v
ar
i
o
u
s
b
as
e
s
o
f
w
av
el
e
t
s
u
c
h
as
:
D
au
b
ech
i
es
(d
b
),
b
i
o
r
t
h
o
g
o
n
al
(b
i
o
r.
)
an
d
s
y
ml
e
t
(s
y
m.
),
s
h
o
w
t
h
a
t
d
e
n
o
i
s
i
n
g
p
ro
ce
s
s
t
h
a
t
u
s
es
i
n
t
h
i
s
m
et
h
o
d
p
r
o
d
u
ces
e
x
t
ra
p
ro
m
i
n
e
n
t
i
ma
g
es
a
n
d
i
m
p
ro
v
ed
v
al
u
es
o
f
PS
N
R
t
h
a
n
o
t
h
er
met
h
o
d
s
.
K
e
y
w
o
r
d
s
:
C
olor
e
d
n
ois
e
De
nois
ing
E
s
ti
mation
P
S
NR
W
a
ve
let
tr
a
ns
f
or
m
Th
i
s
i
s
a
n
o
p
en
a
c
ces
s
a
r
t
i
c
l
e
u
n
d
e
r
t
h
e
CC
B
Y
-
SA
l
i
ce
n
s
e
.
C
or
r
e
s
pon
din
g
A
u
th
or
:
Ya
s
in
Yous
if
Al
-
Aboos
i
,
F
a
c
ult
y
of
E
nginee
r
ing
,
Unive
r
s
it
y
of
M
us
tans
ir
iyah,
B
a
ghda
d,
I
r
a
q
.
E
mail:
a
laboos
iyas
in@gm
a
il
.
c
om
1.
I
NT
RODU
C
T
I
ON
E
f
f
icie
nt
unde
r
wa
ter
i
mage
de
nois
ing
is
a
c
r
it
ica
l
a
s
pe
c
t
f
or
many
a
ppli
c
a
ti
ons
[
1]
.
Unde
r
wa
ter
im
a
ge
s
pr
e
s
e
nt
two
main
pr
oblems
:
li
ght
s
c
a
tt
e
r
ing
that
a
lt
e
r
s
li
ght
pa
th
dir
e
c
ti
on
a
nd
c
olo
r
c
ha
nge
.
T
he
ba
s
ic
pr
oc
e
s
s
e
s
in
unde
r
wa
ter
li
ght
pr
opa
ga
ti
on
a
r
e
s
c
a
tt
e
r
ing
a
nd
a
bs
or
pti
on.
Un
de
r
wa
ter
nois
e
ge
ne
r
a
ll
y
o
r
igi
na
tes
f
r
om
man
-
made
(
e
.
g
.
s
hippi
ng
a
nd
mac
hiner
y
s
ounds
)
a
nd
na
tur
a
l
(
e
.
g.
wind
,
s
e
is
mi
c
a
nd
r
a
in)
s
our
c
e
s
.
Unde
r
wa
ter
nois
e
r
e
duc
e
s
im
a
ge
qua
li
ty
[
1
,
2
]
,
a
nd
de
nois
ing
ha
s
to
be
a
ppli
e
d
to
im
pr
ove
it
[
3
]
.
Unde
r
wa
ter
s
ound
a
tt
e
nua
ti
on
is
de
pe
nde
nt
on
f
r
e
que
nc
y.
C
ons
e
que
ntl
y,
powe
r
s
pe
c
tr
a
l
de
ns
it
y
(
P
S
D)
f
or
a
mbi
e
nt
nois
e
is
de
f
ined
a
s
c
olo
r
e
d
[
4
]
.
M
a
ny
im
a
ge
de
nois
ing
tec
hniques
a
r
e
de
s
c
r
ibed
i
n
[
5
-
9
]
.
A
method
ba
s
e
d
on
a
da
pti
ve
wa
ve
let
with
a
da
pti
ve
thr
e
s
hold
s
e
lec
ti
on
wa
s
s
ugge
s
ted
in
[
5
]
to
ove
r
c
ome
the
unde
r
wa
ter
im
a
ge
de
nois
ing
pr
oblem.
As
s
ume
that
a
n
unde
r
wa
ter
im
a
ge
ha
s
a
s
mall
s
ignal
-
to
-
nois
e
r
a
ti
o
(
S
NR
)
a
nd
im
a
ge
qua
li
ty
is
poor
.
T
he
s
im
ulation
r
e
s
ult
s
s
how
that
the
pr
opos
e
d
met
hod
s
uc
c
e
s
s
f
ull
y
e
li
mi
na
tes
nois
e
,
im
pr
ove
s
the
pe
a
k
S
NR
(
P
S
NR
)
output
of
the
im
a
ge
a
nd
pr
oduc
e
s
a
high
-
q
ua
li
ty
im
a
ge
.
L
ight
is
r
e
pe
a
tedly
de
f
lec
ted
a
nd
r
e
f
l
e
c
ted
by
e
xis
ti
ng
pa
r
ti
c
les
in
the
wa
ter
due
to
the
l
ight
s
c
a
tt
e
r
ing
phe
nomenon,
whic
h
de
g
r
a
de
s
the
vis
ib
il
it
y
a
nd
c
ontr
a
s
t
of
unde
r
wa
ter
im
a
ge
s
.
T
he
r
e
f
o
r
e
,
unde
r
w
a
ter
im
a
ge
s
e
xhibi
t
poor
qua
li
ty
.
T
o
pr
oc
e
s
s
im
a
ge
s
f
ur
ther
,
wa
ve
let
tr
a
ns
f
or
m
a
nd
W
e
be
r
’
s
law
we
r
e
pr
opos
e
d
in
[
8
]
.
F
ir
s
tl
y,
s
e
ve
r
a
l
p
r
e
-
pr
oc
e
s
s
ing
methodologi
e
s
we
r
e
c
onduc
ted
pr
ior
to
wa
ve
let
de
nois
ing
thr
e
s
holdi
ng.
T
he
n,
W
e
be
r
’
s
law
wa
s
us
e
d
f
o
r
im
a
ge
e
nha
nc
e
ment
a
long
with
wa
ve
let
tr
a
ns
f
or
m
.
C
ons
e
que
ntl
y,
th
e
r
e
c
ove
r
e
d
im
a
ge
s
we
r
e
e
nha
nc
e
d
a
nd
the
noi
s
e
leve
l
wa
s
r
e
duc
e
d.
I
n
the
c
ur
r
e
nt
s
tudy,
a
nove
l
im
a
ge
d
e
nois
ing
method
is
p
r
opos
e
d
in
the
p
r
e
s
e
nc
e
of
un
de
r
wa
ter
nois
e
us
ing
a
pr
e
-
whitening
f
il
ter
a
nd
dis
c
r
e
te
wa
v
e
let
tr
a
ns
f
or
m
(
DW
T
)
with
s
ingl
e
-
leve
l
e
s
ti
mation.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
S
S
N
:
1693
-
6930
T
E
L
KO
M
NI
KA
T
e
lec
omm
un
C
omput
E
l
C
ontr
o
l
,
Vol.
18
,
No
.
3
,
J
une
2020:
14
3
9
–
14
4
6
1440
2.
AM
B
I
E
NT
NOI
S
E
CHAR
AC
T
E
RI
S
T
I
CS
T
he
c
ha
r
a
c
ter
is
ti
c
s
of
unde
r
wa
ter
nois
e
in
s
e
a
s
h
a
ve
be
e
n
dis
c
us
s
e
d
e
xtens
ively
[
10]
.
S
uc
h
nois
e
ha
s
f
our
c
omponents
:
tu
r
bulenc
e
,
s
hippi
ng
,
wind
a
nd
th
e
r
mal
nois
e
s
.
E
a
c
h
c
omponent
oc
c
upies
a
c
e
r
tain
f
r
e
que
nc
y
ba
nd
of
s
pe
c
tr
um.
E
a
c
h
nois
e
s
our
c
e
is
domi
na
nt
in
c
e
r
tain
f
r
e
que
nc
y
ba
nds
,
a
s
ind
ica
ted
in
T
a
ble
1.
T
he
P
S
D
o
f
e
a
c
h
c
o
mponent
is
e
xp
r
e
s
s
e
d
a
s
[
1,
2
,
11
];
(
1)
(2
)
(
3)
(
4)
whe
r
e
f
r
e
pr
e
s
e
nts
the
f
r
e
que
nc
y
in
KH
z
.
T
he
r
e
f
or
e
,
the
tot
a
l
P
S
D
of
unde
r
wa
ter
nois
e
f
o
r
a
given
f
r
e
que
nc
y
f
(
kHz
)
is
;
(5
)
T
a
ble
1.
Unde
r
wa
ter
nois
e
ba
nd
B
a
nd
T
ype
0.1H
z
-
10H
z
T
ur
bul
e
nc
e
noi
s
e
10H
z
-
200Hz
S
hi
ppi
ng nois
e
0.2kHz
-
100kHz
W
in
d
noi
s
e
A
bove
kH
z
T
he
r
ma
l
noi
s
e
3.
I
M
AGE
M
ODE
L
I
N
P
RE
S
E
NC
E
OF
COL
OR
E
D
NOI
S
E
Nois
e
int
e
r
f
e
r
e
nc
e
is
a
c
omm
on
pr
oblem
in
digi
tal
c
omm
unica
ti
on
a
nd
im
a
ge
p
r
oc
e
s
s
ing.
An
unde
r
wa
ter
nois
e
model
f
or
im
a
g
e
de
nois
ing
in
a
n
a
ddit
ive
c
olo
r
e
d
nois
e
c
ha
nne
l
is
p
r
e
s
e
nte
d
in
thi
s
s
e
c
ti
on.
Nume
r
ous
a
ppli
c
a
ti
ons
a
s
s
ume
that
a
r
e
c
e
ived
im
a
ge
c
a
n
be
e
xpr
e
s
s
e
d
a
s
f
oll
ows
;
(6
)
whe
r
e
(
)
is
the
or
igi
na
l
im
a
ge
a
nd
(
)
de
notes
unde
r
wa
ter
nois
e
.
He
nc
e
,
de
nois
ing
a
im
s
to
e
li
mi
na
te
the
c
or
r
upt
ion
de
g
r
e
e
of
(
)
c
a
us
e
d
by
(
)
.
T
he
powe
r
s
pe
c
tr
um
a
nd
a
utocor
r
e
lation
o
f
a
ddit
ive
white
Ga
us
s
ian
nois
e
(
AW
GN
)
a
r
e
e
xpr
e
s
s
e
d
a
s
[
12
]
;
(7
)
(
8)
T
he
P
S
D
of
AW
GN
r
e
mains
c
ons
tant
a
c
r
os
s
t
he
e
nti
r
e
f
r
e
que
nc
y
r
a
nge
,
in
whic
h
a
ll
r
a
nge
s
of
f
r
e
que
nc
ies
ha
ve
a
magnitude
of
σ
v
2
.
T
he
p
r
oba
bil
it
y
dis
tr
ibut
ion
f
unc
ti
on
(
)
f
or
AW
GN
is
s
pe
c
if
ied
by
[
13
]
;
(9
)
whe
r
e
r
e
pr
e
s
e
nts
the
s
tanda
r
d
de
viation.
W
it
h
r
e
ga
r
d
to
a
utocor
r
e
lation
f
unc
ti
ons
,
the
de
lt
a
f
u
nc
ti
on
indi
c
a
tes
that
a
djac
e
nt
s
a
mpl
e
s
a
r
e
indepe
nde
nt.
T
he
r
e
f
or
e
,
obs
e
r
ve
d
s
a
mpl
e
s
a
r
e
c
ons
ider
e
d
inde
pe
nde
nt
a
nd
identica
ll
y
dis
tr
ibut
e
d.
Unde
r
wa
ter
nois
e
is
de
pe
n
de
nt
on
f
r
e
que
nc
y
[
14
,
15
]
;
a
nd
it
is
s
uit
a
bly
modelled
a
s
c
olor
e
d
nois
e
[
1
,
2
,
16
]
.
T
he
P
S
D
of
c
olo
r
e
d
noi
s
e
is
de
f
ined
a
s
[
17
,
18
]
;
(
10)
Evaluation Warning : The document was created with Spire.PDF for Python.
T
E
L
KO
M
NI
KA
T
e
lec
omm
un
C
omput
E
l
C
ontr
o
l
I
mage
de
nos
ing
in
unde
r
w
ater
ac
ous
ti
c
nois
e
us
ing
dis
c
r
e
te…
(
Y
as
in
Y
ous
if
A
l
-
A
boos
i)
1441
How
e
ve
r
,
the
[
]
of
c
olou
r
e
d
nois
e
is
not
li
ke
a
de
lt
a
f
unc
ti
on,
but,
it
is
take
s
the
f
o
r
mul
a
of
a
(
)
f
unc
ti
on
[
12
,
17
]
.
I
n
c
ont
r
a
s
t
to
AW
GN
,
nois
e
s
a
mpl
e
s
a
r
e
c
or
r
e
late
d
[
18
]
.
4.
I
M
AGE
DE
NOI
S
I
NG
W
a
ve
lets
a
r
e
us
e
d
in
im
a
ge
pr
oc
e
s
s
ing
f
or
s
a
mpl
e
e
dge
de
tec
ti
on,
wa
ter
mar
king
,
c
ompr
e
s
s
ion,
de
nois
ing
a
nd
c
oding
of
int
e
r
e
s
ti
ng
f
e
a
tu
r
e
s
f
or
s
u
bs
e
que
nt
c
las
s
if
ica
ti
on
[
19
,
2
0
]
.
T
he
f
ol
lowing
s
ubs
e
c
ti
ons
dis
c
us
s
im
a
ge
de
nois
ing
by
thr
e
s
holdi
ng
the
DW
T
c
oe
f
f
icie
nts
.
4.
1.
DWT
of
an
im
age
d
at
a
An
im
a
ge
is
p
r
e
s
e
nted
a
s
a
2D
a
r
r
a
y
of
c
oe
f
f
ic
ients
.
E
a
c
h
c
oe
f
f
icie
nt
r
e
pr
e
s
e
nts
the
br
ight
ne
s
s
de
gr
e
e
at
that
point
.
M
os
t
he
r
ba
l
photogr
a
phs
e
xhibi
t
s
moot
h
c
olo
r
a
ti
on
va
r
iations
with
e
xc
e
ll
e
nt
de
tails
r
e
pr
e
s
e
nted
a
s
s
ha
r
p
e
dge
s
a
mong
e
a
s
y
ve
r
s
ions
.
C
lea
n
va
r
iations
in
c
olo
r
a
ti
on
c
a
n
b
e
s
tr
ictly
labe
ll
e
d
a
s
low
-
f
r
e
que
nc
y
ve
r
s
ions
,
whe
r
e
a
s
point
y
va
r
iations
c
a
n
be
labe
ll
e
d
a
s
e
xc
e
s
s
ive
-
f
r
e
que
nc
y
ve
r
s
ions
.
T
he
low
-
f
r
e
que
nc
y
c
omponents
(
i.
e
.
s
moot
h
ve
r
s
ions
)
e
s
tablis
h
the
ba
s
e
of
a
photogr
a
ph,
whe
r
e
a
s
the
e
xc
e
s
s
ive
-
f
r
e
que
nc
y
c
omponents
(
i.
e
.
the
e
dge
s
that
p
r
ovide
the
de
tails
)
a
r
e
upload
e
d
upon
the
low
-
f
r
e
que
nc
y
c
omponents
to
r
e
f
ine
the
im
a
ge
,
ther
e
by
p
r
oduc
ing
a
n
in
-
de
pth
im
a
ge
.
T
he
r
e
f
o
r
e
,
the
e
a
s
y
ve
r
s
ions
a
r
e
mor
e
im
por
tant
than
the
de
tails
.
Nu
mer
ous
methods
c
a
n
be
us
e
d
to
dis
ti
nguis
h
be
tw
e
e
n
e
a
s
y
va
r
iations
a
nd
photogr
a
ph
inf
o
r
mation.
One
e
xa
mpl
e
of
thes
e
methods
is
pictur
e
de
c
ompos
it
ion
via
DW
T
r
e
modeling
.
T
he
di
f
f
e
r
e
nt
de
c
ompos
it
ion
leve
ls
of
DW
T
a
r
e
s
hown
in
F
igur
e
1.
F
igur
e
1
.
DW
T
De
c
ompos
it
ion
leve
ls
4.
2.
T
h
e
I
n
ve
r
s
e
DWT
of
an
im
age
Dif
f
e
r
e
nt
c
las
s
e
s
of
da
ta
a
r
e
c
oll
e
c
ted
in
to
a
r
e
c
ons
tr
uc
ted
im
a
ge
by
us
ing
r
e
ve
r
s
e
wa
ve
let
tr
a
ns
f
or
m.
A
pa
ir
of
high
-
a
nd
low
-
pa
s
s
f
il
ter
s
is
a
ls
o
us
e
d
du
r
ing
the
r
e
c
ons
tr
uc
ti
on
p
r
oc
e
s
s
.
T
hi
s
pa
ir
of
f
il
ter
s
is
r
e
f
e
r
r
e
d
to
a
s
the
s
ynthes
is
f
il
te
r
p
a
ir
.
T
he
f
i
lt
e
r
ing
pr
oc
e
dur
e
is
s
im
ply
the
op
pos
it
e
of
tr
a
ns
f
or
mation;
that
is
,
the
p
r
oc
e
dur
e
s
tar
ts
f
r
om
t
he
highes
t
leve
l.
T
he
f
il
ter
s
a
r
e
f
i
r
s
tl
y
a
ppli
e
d
c
olum
n
-
wis
e
a
nd
then
r
ow
-
wis
e
leve
l
by
leve
l
unt
il
the
lowe
s
t
leve
l
is
r
e
a
c
he
d.
5.
P
ROP
OS
E
D
M
E
T
HO
D
I
n
thi
s
pa
pe
r
,
the
DW
T
is
us
e
d
f
o
r
the
tr
a
ns
f
or
mation
of
im
a
ge
in
the
pr
oc
e
s
s
of
de
nois
ing.
T
he
a
dva
ntage
s
of
us
e
d
mul
ti
-
leve
l
th
r
e
s
hold
e
s
ti
mation
in
de
nois
ing
p
r
oc
e
s
s
to
r
e
duc
e
the
r
e
quir
e
d
of
the
us
e
of
the
pr
e
-
whitening
s
tage
in
c
a
s
e
of
us
ing
s
ingl
e
leve
l
thr
e
s
hold
e
s
ti
mation
[
21
]
.
T
he
f
ol
lowi
ng
s
teps
de
s
c
r
ibe
the
im
a
ge
de
nois
ing
pr
oc
e
dur
e
.
a)
T
he
DW
T
o
f
a
no
is
y
im
a
ge
is
c
ompu
ted.
T
he
W
T
is
(
ti
me
-
f
r
e
que
nc
y
dis
tr
ibut
ion)
that
us
e
d
to
de
c
ompos
e
s
ignal
int
o
f
a
mi
ly
of
f
unc
ti
ons
loca
li
z
e
in
f
r
e
que
nc
y
a
nd
ti
m
e
.
T
he
C
W
T
c
a
n
be
de
s
c
r
ibed
a
s
;
(
11
)
whe
r
e
is
s
hif
ti
ng
in
ti
me
a
nd
a
is
s
c
a
le
f
a
c
tor
o
r
dil
a
ti
on
f
a
c
tor
a
nd
h(
t)
is
r
e
pr
e
s
e
nt
ba
s
is
f
unc
ti
on.
Da
ube
c
hies
,
c
oif
let,
s
yml
e
t
,
a
nd
b
ior
thogonal
r
e
p
r
e
s
e
nt
e
xa
mpl
e
s
of
f
unc
ti
ons
us
e
d
in
C
W
T
a
s
s
ho
wn
in
F
igur
e
2.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
S
S
N
:
1693
-
6930
T
E
L
KO
M
NI
KA
T
e
lec
omm
un
C
omput
E
l
C
ontr
o
l
,
Vol.
18
,
No
.
3
,
J
une
2020:
14
3
9
–
14
4
6
1442
(
a
)
(
b)
(
c
)
(
d)
F
igur
e
2
.
S
ome
ba
s
ic
f
unc
ti
ons
us
e
d
i
n
W
T
a
r
e
:
(
a
)
Ha
a
r
,
(
b
)
S
yml
e
t
6
,
(
c
)
Da
ube
c
hies
6
,
(
d)
B
ior
t
hogona
l
1.
5
[
22
]
b)
Af
ter
the
DW
T
r
e
pr
e
s
e
ntation
done
,
de
-
nois
ing
is
done
us
ing
s
of
t
-
t
hr
e
s
holdi
ng
by
modi
f
ied
univ
e
r
s
a
l
thr
e
s
hold
e
s
ti
mation
(
M
UT
E
)
.
P
r
ovidi
ng
a
mbi
e
n
t
nois
e
is
a
c
olo
r
e
d,
a
th
r
e
s
hold
de
pe
nde
nt
on
leve
l
a
ppli
e
d
to
e
a
c
h
leve
l
of
f
r
e
que
nc
y
wa
s
pr
opos
e
d
in
[
23
]
.
T
he
va
lue
of
thr
e
s
hold
a
pp
li
e
d
to
the
c
oe
f
f
ic
ients
of
e
s
ti
mate
d
ti
me
-
f
r
e
que
nc
y
us
ing
M
UT
E
[
23
]
is
e
xpr
e
s
s
e
d
a
s
;
(
12)
whe
r
e
N
is
length
of
s
ignal,
,
is
nois
e
e
s
ti
mate
d
s
t
a
nda
r
d
de
viation
f
o
r
leve
l
k,
a
nd
c
is
the
(
modi
f
ie
d
univer
s
a
l
thr
e
s
hold
f
a
c
tor
)
0
<
<
1
.
T
he
s
tanda
r
d
de
via
ti
o
n
f
or
nois
e
a
t
e
a
c
h
leve
l
is
;
,
=
(
|
(
,
)
|
)
0
.
6745
(
13)
whe
r
e
(
,
)
r
e
pr
e
s
e
nts
a
ll
the
c
oe
f
f
icie
nts
f
o
r
f
r
e
que
nc
y
leve
l
k
[
24
]
.
T
he
va
lue
o
f
thr
e
s
hold
is
us
e
d
to
r
e
movi
ng
the
nois
e
a
nd
a
ls
o
f
or
e
f
f
icie
nt
r
e
c
ove
r
or
igi
na
l
s
ignal.
T
he
thr
e
s
hold
f
a
c
tor
c
is
us
e
d
to
im
p
r
ove
f
ur
ther
pe
r
f
or
manc
e
o
f
de
nois
ing
[
25
]
.
T
he
va
lue
o
f
c
is
c
a
lcula
t
e
d
gr
a
dua
ll
y
by
incr
e
ment
it
o
f
0
.
1
f
or
e
a
c
h
leve
l
to
ob
tain
the
be
s
t
r
e
s
ult
s
a
t
highes
t
P
S
NR
.
c)
A
f
ter
va
lues
of
thr
e
s
hold
is
de
ter
m
ined
f
o
r
a
ll
c
omponents
,
the
c
omponents
r
e
pr
e
s
e
ntations
of
ti
me
-
f
r
e
que
nc
y
a
f
ter
ha
r
d
-
thr
e
s
holdi
ng
a
r
e
;
(
,
)
=
{
(
,
)
|
(
,
)
|
>
0
|
(
,
)
|
≤
(1
4)
a
nd
the
c
omponents
a
f
ter
s
of
t
-
thr
e
s
holdi
ng
a
r
e
;
Evaluation Warning : The document was created with Spire.PDF for Python.
T
E
L
KO
M
NI
KA
T
e
lec
omm
un
C
omput
E
l
C
ontr
o
l
I
mage
de
nos
ing
in
unde
r
w
ater
ac
ous
ti
c
nois
e
us
ing
dis
c
r
e
te…
(
Y
as
in
Y
ous
if
A
l
-
A
boos
i)
1443
(
15)
whe
r
e
de
notes
the
thr
e
s
hold
va
lue
in
leve
l
k.
d)
T
he
im
a
ge
is
r
e
c
ons
tr
uc
ted
by
a
pplyi
ng
inver
s
e
DW
T
to
obtain
the
de
nois
e
d
im
a
ge
.
F
igu
r
e
3
s
h
ows
the
da
ta
f
low
d
iagr
a
m
o
f
the
im
a
ge
de
nois
ing
pr
oc
e
s
s
.
T
he
I
W
T
is
e
xpr
e
s
s
e
d
a
s
;
(
)
=
∫
∫
(
,
)
1
√
ℎ
(
−
)
2
∞
−
∞
∞
0
(
16)
F
igur
e
3
.
Da
ta
f
low
diagr
a
m
of
im
a
ge
de
nois
ing
us
ing
leve
l
-
de
pe
nde
nt
e
s
ti
mation
dis
c
r
e
te
wa
ve
let
tr
a
ns
f
or
ms
6.
P
E
RF
ORM
AN
CE
M
E
ASURE
S
C
omm
on
mea
s
ur
e
ment
pa
r
a
mete
r
s
f
o
r
i
mage
r
e
l
iabili
ty
include
mea
n
a
bs
olut
e
e
r
r
o
r
,
nor
malize
d
M
S
E
(
NM
S
E
)
,
P
S
NR
,
a
nd
M
S
E
[
26
]
.
An
S
NR
ove
r
40
dB
pr
ovides
e
xc
e
ll
e
nt
im
a
ge
qua
li
ty
that
is
c
los
e
to
that
of
the
or
igi
na
l
im
a
ge
;
a
n
S
NR
of
30
-
40
d
B
typi
c
a
ll
y
pr
oduc
e
s
good
im
a
ge
qua
li
ty
with
a
c
c
e
ptable
dis
tor
ti
on;
a
n
S
NR
of
20
-
30
dB
pr
e
s
e
nts
poor
im
a
ge
qua
li
ty;
a
n
S
NR
be
low
20
dB
ge
ne
r
a
tes
a
n
una
c
c
e
ptable
im
a
ge
[
27
]
.
T
he
c
a
lcula
ti
on
methods
of
P
S
NR
a
nd
NM
S
E
[
28
]
a
r
e
p
r
e
s
e
nted
a
s
f
oll
ows
:
(
17)
whe
r
e
M
S
E
is
the
M
S
E
be
twe
e
n
the
or
igi
na
l
im
a
g
e
(
)
a
nd
the
de
nois
e
d
im
a
ge
(
̂
)
with
s
ize
M
×
N:
(
18)
7.
RE
S
UL
T
S
AN
D
DI
S
CU
S
S
I
ON
M
AT
L
AB
is
us
e
d
a
s
the
e
xpe
r
im
e
ntal
tool
f
or
s
im
ulation,
a
nd
s
im
ulation
e
xpe
r
im
e
nts
a
r
e
pe
r
f
or
med
on
a
diver
im
a
ge
to
c
on
f
ir
m
the
va
li
dit
y
of
the
a
lgor
it
hm.
T
he
s
im
ulations
a
r
e
a
c
hieve
d
a
t
P
S
NR
r
a
nging
f
r
om
30
dB
to
60
dB
by
c
ha
nging
nois
e
powe
r
f
r
o
m
0
dB
t
o
15
dB
.
T
he
a
ppli
e
d
o
r
de
r
of
the
whiteni
ng
f
il
ter
is
10.
Di
f
f
e
r
e
nt
de
nois
ing
wa
ve
let
bias
e
s
(
i.
e
.
Da
ube
c
hies
,
bior
thogonal
1.
5
,
a
nd
s
yml
e
t)
a
r
e
t
e
s
ted
on
a
n
im
a
ge
with
unde
r
wa
ter
nois
e
via
nu
mer
ica
l
s
i
mul
a
ti
on.
As
s
hown
in
F
igu
r
e
4,
s
of
t
th
r
e
s
holdi
ng
a
nd
f
our
de
c
ompos
it
ion
leve
ls
a
r
e
us
e
d.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
S
S
N
:
1693
-
6930
T
E
L
KO
M
NI
KA
T
e
lec
omm
un
C
omput
E
l
C
ontr
o
l
,
Vol.
18
,
No
.
3
,
J
une
2020:
14
3
9
–
14
4
6
1444
Bi
a
s
t
y
pe
N
o
i
s
y
i
m
a
g
e
De
-
no
i
s
e
i
m
a
g
e
P
SN
R
(dB)
Sy
m4
4
5
.
7
d
B
Sy
m4
2
9
.
0
1
d
B
5
4
5
.
7
7
d
B
5
2
8
.
9
6
d
B
Bi
o
rt
h
o
g
o
n
a
l
1
.
3
4
5
.
2
7
d
B
Bi
o
rt
h
o
g
o
n
a
l
1
.
3
2
8
.
9
7
d
B
F
igur
e
4
.
S
im
ulation
r
e
s
ult
s
on
d
iver
im
a
ge
us
ing
dif
f
e
r
e
nt
wa
ve
let
b
ias
e
s
Evaluation Warning : The document was created with Spire.PDF for Python.
T
E
L
KO
M
NI
KA
T
e
lec
omm
un
C
omput
E
l
C
ontr
o
l
I
mage
de
nos
ing
in
unde
r
w
ater
ac
ous
ti
c
nois
e
us
ing
dis
c
r
e
te…
(
Y
as
in
Y
ous
if
A
l
-
A
boos
i)
1445
T
a
bles
2,
3,
a
nd
4
s
how
the
pe
r
f
o
r
manc
e
of
the
p
r
opos
e
d
method
on
va
r
ious
nois
e
powe
r
ba
s
e
d
on
the
Da
ube
c
hies
,
s
yml
e
t
,
a
nd
bior
thogonal
wa
ve
let
bias
e
s
,
r
e
s
pe
c
ti
ve
ly.
T
he
P
S
NR
a
nd
M
S
E
va
lues
a
r
e
c
a
lcula
ted
ba
s
e
d
on
e
a
c
h
nois
e
powe
r
va
lue.
T
a
ble
2
.
P
e
r
f
o
r
manc
e
r
e
s
ult
s
of
P
S
NR
a
nd
M
S
E
o
n
diver
im
a
ge
ba
s
e
d
on
Da
ube
c
hies
wa
ve
let
bias
N
oi
s
e
pow
e
r
(
db)
P
S
N
R
M
S
E
0
59.35
0.0756
3
55.57
0.2161
5
52.3
0.5226
10
42.9653
0.5062
15
33.2530
0.8279
T
a
ble
3
.
P
e
r
f
o
r
manc
e
r
e
s
ult
s
of
P
S
NR
a
nd
M
S
E
o
n
diver
im
a
ge
ba
s
e
d
on
s
yml
e
t
wa
ve
let
bias
N
oi
s
e
pow
e
r
(
db)
P
S
N
R
M
S
E
0
61.066
0.0759
3
55.48
0.2948
5
52.032
0.5509
10
43.33
0.4420
15
35.0591
0.392
T
a
ble
4
.
P
e
r
f
o
r
manc
e
r
e
s
ult
s
of
P
S
NR
a
nd
M
S
E
o
n
diver
im
a
ge
ba
s
e
d
on
bior
thogonal
wa
ve
let
bias
N
oi
s
e
pow
e
r
(
db)
P
S
N
R
M
S
E
0
59.7560
0.0773
3
55.76
0.321
5
52.867
0.5240
10
43.7970
0.767
15
35.125
0.8604
8.
CONC
L
USI
ON
Unde
r
wa
ter
nois
e
is
mainly
c
ha
r
a
c
ter
ize
d
a
s
non
-
white
a
nd
non
-
Ga
us
s
ian
nois
e
.
T
he
r
e
f
or
e
,
tr
a
dit
ional
methods
us
e
d
f
or
im
a
ge
de
nois
ing
us
ing
wa
ve
let
tr
a
ns
f
or
m
unde
r
wa
ter
a
r
e
inef
f
icie
nt
be
c
a
us
e
thes
e
methods
us
e
mul
ti
-
leve
l
e
s
ti
mation
dis
c
r
e
te
wa
ve
let
tr
a
ns
f
or
m
f
or
nois
e
va
r
ianc
e
.
How
e
ve
r
,
nois
e
va
r
ianc
e
a
t
e
a
c
h
leve
l
s
hould
be
indepe
nde
ntl
y
e
s
ti
mate
d
in
c
olor
e
d
nois
e
.
T
he
wa
ve
let
de
nois
ing
method
c
a
n
be
e
f
f
icie
ntl
y
us
e
d
with
P
S
NR
a
nd
M
S
E
c
o
mpar
e
d
to
other
method
us
e
d
a
p
r
e
-
whitening
f
i
lt
e
r
that
c
onve
r
ts
un
de
r
wa
ter
nois
e
to
white
nois
e
,
a
s
de
mo
ns
tr
a
ted
by
the
r
e
s
ult
s
.
RE
F
E
RE
NC
E
S
[
1]
G
.
Bu
rro
w
e
s
an
d
J
.
Y
.
K
h
an
,
"
Sh
o
rt
-
ran
g
e
u
n
d
erw
a
t
er
aco
u
s
t
i
c
co
mmu
n
i
cat
i
o
n
n
et
w
o
r
k
s
,
au
t
o
n
o
mo
u
s
u
n
d
erw
at
er
v
eh
i
cl
e
s
,"
INTE
CH
O
p
e
n
A
cce
s
s
P
u
b
l
i
s
h
er
,
p
p
.
1
7
3
-
1
9
8
,
2011.
[
2]
T
.
Mel
o
d
i
a,
H
.
K
u
l
h
an
d
j
i
a
n
,
L
-
C.
K
u
o
,
an
d
E
.
D
em
i
ro
rs
,
"
A
d
v
a
n
ces
i
n
u
n
d
erw
a
t
er
aco
u
s
t
i
c
n
et
w
o
r
k
i
n
g
,
"
M
o
b
i
l
e
A
d
H
o
c
Net
w
o
r
k
i
n
g
:
Cu
t
t
i
n
g
E
d
g
e
D
i
r
ect
i
o
n
s
,
p
p
.
8
0
4
-
8
5
2
,
2
0
1
3
.
[
3]
H
.
O
l
k
k
o
n
e
n
,
"
D
i
s
cre
t
e
w
a
v
el
e
t
t
ran
s
fo
rm
s
:
A
l
g
o
r
i
t
h
ms
an
d
ap
p
l
i
cat
i
o
n
s
,
"
INTE
CH
O
p
e
n
,
2
0
1
1
.
[
4]
A
.
Z
.
Sh
a'ameri
,
Y
.
Y
.
A
l
-
A
b
o
o
s
i
,
a
n
d
N
.
H
.
H
.
K
h
a
mi
s
,
"
U
n
d
erw
a
t
er
aco
u
s
t
i
c
n
o
i
s
e
c
h
aract
er
i
s
t
i
c
s
o
f
s
h
a
l
l
o
w
w
at
er
i
n
t
r
o
p
i
cal
s
ea
s
,
"
In
t
e
r
n
a
t
i
o
n
a
l
Co
n
f
er
e
n
ce
o
n
Co
m
p
u
t
e
r
a
n
d
Co
m
m
u
n
i
ca
t
i
o
n
E
n
g
i
n
ee
r
i
n
g
,
p
p
.
8
0
-
83
,
2
0
1
4
.
[
5]
P.
A
.
J
ai
s
w
al
an
d
V
.
B.
Pad
o
l
e,
"
Imp
l
eme
n
t
a
t
i
o
n
o
f
an
i
mp
ro
v
ed
al
g
o
r
i
t
h
m
fo
r
u
n
d
erw
a
t
er
i
mag
e
en
h
a
n
cemen
t
an
d
d
en
o
i
s
i
ng,
"
In
t
e
r
n
a
t
i
o
n
a
l
A
d
va
n
ced
R
es
e
a
r
c
h
Jo
u
r
n
a
l
i
n
S
c
i
en
ce,
E
n
g
i
n
e
er
i
n
g
a
n
d
Tech
n
o
l
o
g
y
,
v
o
l
.
3
,
n
o
.
4
,
p
p
.
1
0
7
-
1
1
1
,
2
0
1
6
.
[
6]
L
.
Fei
an
d
W
.
Y
i
n
g
y
i
n
g
,
"
T
h
e
res
earc
h
o
f
u
n
d
er
w
at
er
i
mag
e
d
e
-
n
o
i
s
i
n
g
met
h
o
d
b
as
ed
o
n
ad
a
p
t
i
v
e
w
a
v
e
l
et
t
ran
s
fo
rm,
"
2
6
t
h
Ch
i
n
e
s
e
Co
n
t
r
o
l
a
n
d
D
ec
i
s
i
o
n
Co
n
f
er
e
n
ce
,
pp
.
2
5
2
1
-
2
5
2
5
,
2
0
1
4
.
[
7]
R.
Sat
h
y
a,
M.
Bh
arat
h
i
,
an
d
G
.
D
h
i
v
y
as
r
i
,
"
U
n
d
er
w
at
er
i
mag
e
en
h
an
c
emen
t
b
y
d
ark
ch
a
n
n
e
l
p
ri
o
r,
"
2
nd
In
t
e
r
n
a
t
i
o
n
a
l
Co
n
f
er
e
n
ce
o
n
E
l
ec
t
r
o
n
i
cs
a
n
d
Co
m
m
u
n
i
c
a
t
i
o
n
S
ys
t
em
s
(ICE
C
S
)
,
p
p
.
1
1
1
9
-
1
1
2
3
,
2
0
1
5
.
[
8]
S.
J
i
an
an
d
W
.
W
en
,
"
St
u
d
y
o
n
u
n
d
er
w
at
er
i
ma
g
e
d
en
o
i
s
i
n
g
al
g
o
ri
t
h
m
b
a
s
ed
o
n
w
a
v
el
e
t
t
ran
s
fo
rm
,
"
Jo
u
r
n
a
l
o
f
P
h
y
s
i
c
s
:
Co
n
f
er
e
n
ce
S
e
r
i
e
s
,
v
o
l
.
8
0
6
,
p
p
.
1
-
9
,
2
0
1
7
.
[
9]
R.
Sch
et
t
i
n
i
a
n
d
S.
C
o
rch
s
,
"
U
n
d
erw
a
t
er
i
ma
g
e
p
ro
ce
s
s
i
n
g
:
s
t
a
t
e
o
f
t
h
e
art
o
f
re
s
t
o
rat
i
o
n
an
d
i
ma
g
e
e
n
h
a
n
cem
en
t
met
h
o
d
s
,
"
E
U
R
A
S
I
P
Jo
u
r
n
a
l
o
n
A
d
v
a
n
ce
s
i
n
S
i
g
n
a
l
P
r
o
ces
s
i
n
g
,
v
o
l
.
2
0
1
0
,
p
p
.
1
-
14
,
2
0
1
0
.
[
10]
R.
J
.
U
ri
ck
,
"
Pri
n
c
i
p
l
es
o
f
u
n
d
erw
a
t
er
s
o
u
n
d
,
"
M
cG
r
a
w
-
H
i
l
l
,
1
9
8
3
.
[
11]
Y
.
Y
.
A
l
-
A
b
o
o
s
i
,
A
.
K
an
aa,
A
.
Z
.
Sh
a'ameri
,
an
d
H
.
A
.
A
b
d
u
a
l
n
a
b
i
,
"
D
i
u
r
n
al
v
ari
a
b
i
l
i
t
y
o
f
u
n
d
er
w
at
er
ac
o
u
s
t
i
c
n
o
i
s
e
ch
ar
act
er
i
s
t
i
c
s
i
n
s
h
a
l
l
o
w
w
at
e
r,
"
T
E
LK
O
M
NIK
A
Tel
eco
m
m
u
n
i
c
a
t
i
o
n
,
Co
m
p
u
t
i
n
g
,
E
l
ec
t
r
o
n
i
cs
a
n
d
Co
n
t
r
o
l
,
v
o
l
.
1
5
,
n
o
.
1
,
p
p
.
3
1
4
-
3
2
1
,
2
0
1
7
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
S
S
N
:
1693
-
6930
T
E
L
KO
M
NI
KA
T
e
lec
omm
un
C
omput
E
l
C
ontr
o
l
,
Vol.
18
,
No
.
3
,
J
une
2020:
14
3
9
–
14
4
6
1446
[
12]
A
.
V
.
O
p
p
en
h
ei
m
an
d
G
.
C.
V
erg
h
es
e,
"
Si
g
n
al
s
,
s
y
s
t
em
s
,
an
d
i
n
fere
n
ce,
"
M
IT
O
p
e
n
Co
u
r
s
eW
a
r
e
,
2
0
1
0
.
[
13]
S.
M.
K
ay
,
"
Fu
n
d
amen
t
al
s
o
f
s
t
at
i
s
t
i
ca
l
s
i
g
n
a
l
p
r
o
ces
s
i
n
g
,
v
o
l
.
II
:
D
et
ec
t
i
o
n
t
h
e
o
ry
,
"
P
r
e
n
t
i
ce
H
a
l
l
,
1
9
9
8
.
[
14]
R.
J
.
U
ri
c
k
,
"
A
mb
i
en
t
n
o
i
s
e
i
n
t
h
e
s
ea,
"
U
n
d
e
r
s
e
a
W
a
r
f
a
r
e
Tec
h
n
o
l
o
g
y
O
f
f
i
ce
Na
v
a
l
S
ea
S
ys
t
e
m
s
C
o
m
m
a
n
d
D
ep
a
r
t
m
e
n
t
o
f
Th
e
Na
vy
W
a
s
h
i
n
g
t
o
n
,
1
9
8
4
.
[
15]
M.
St
o
j
a
n
o
v
i
c
a
n
d
J
.
Prei
s
i
g
,
"
U
n
d
er
w
at
er
ac
o
u
s
t
i
c
co
mmu
n
i
cat
i
o
n
ch
a
n
n
e
l
s
:
Pro
p
a
g
a
t
i
o
n
mo
d
el
s
an
d
s
t
at
i
s
t
i
cal
ch
aract
er
i
zat
i
o
n
,
"
IE
E
E
C
o
m
m
u
n
i
c
a
t
i
o
n
s
M
a
g
a
z
i
n
e
,
v
o
l
.
4
7
,
n
o
.
1
,
p
p
.
8
4
-
8
9
,
2
0
0
9
.
[
16]
M.
Ch
i
t
re,
J
.
Po
t
t
er,
an
d
O
.
S.
H
en
g
,
"
U
n
d
er
w
at
er
ac
o
u
s
t
i
c
ch
an
n
el
c
h
aract
er
i
s
a
t
i
o
n
f
o
r
med
i
u
m
-
ra
n
g
e
s
h
al
l
o
w
w
at
er
co
mm
u
n
i
cat
i
o
n
s
,
"
O
ce
a
n
s
'
0
4
M
TS
/
IE
E
E
Tec
h
no
-
O
cea
n
,
p
p
.
4
0
-
45
,
2
0
0
4
.
[
17]
M.
L
i
an
d
.
W
.
Z
h
ao
,
"
O
n
1
/
f
N
o
i
s
e,
"
H
i
n
d
a
wi
P
u
b
l
i
s
h
i
n
g
Co
r
p
o
r
a
t
i
o
n
M
a
t
h
e
m
a
t
i
c
a
l
P
r
o
b
l
em
s
i
n
E
n
g
i
n
e
er
i
n
g
,
vol.
2
0
1
2
,
p
p
.
1
-
2
3
,
2
0
1
2
.
[
18]
C.
W
.
T
h
erri
en
,
"
D
i
s
cret
e
ran
d
o
m
s
i
g
n
al
s
an
d
s
t
a
t
i
s
t
i
cal
s
i
g
n
a
l
p
ro
ces
s
i
n
g
,
"
P
r
e
n
t
i
ce
H
a
l
l
S
i
g
n
a
l
P
r
o
ces
s
i
n
g
S
e
r
i
es
,
1
9
9
2
.
[
19]
M.
Mi
s
i
t
i
,
Y
.
Mi
s
i
t
i
,
G
.
O
p
p
en
h
e
i
m,
an
d
J
.
M.
Po
g
g
i
,
"
W
av
el
et
t
o
o
l
b
o
x
,
"
Th
e
M
a
t
h
W
o
r
k
s
In
c
.,
Na
t
i
ck,
M
A
,
v
o
l
.
1
5
,
p
.
2
1
,
1
9
9
6
.
[
20]
D
.
Bal
ean
u
,
"
A
d
v
an
ce
s
i
n
w
av
e
l
et
t
h
eo
r
y
an
d
t
h
e
i
r
ap
p
l
i
cat
i
o
n
s
i
n
en
g
i
n
eer
i
n
g
,
p
h
y
s
i
cs
an
d
t
ech
n
o
l
o
g
y
,
"
INTE
CH
O
p
e
n
,
p
p
.
3
5
3
-
3
7
1
,
2
0
1
2
.
[
21]
Y
.
Y
.
A
l
-
A
b
o
o
s
i
an
d
A
.
Z
.
Sh
a'ameri
,
"
Imp
ro
v
ed
u
n
d
er
w
at
er
s
i
g
n
a
l
d
e
t
ect
i
o
n
u
s
i
n
g
effi
c
i
en
t
t
i
me
-
fre
q
u
e
n
cy
de
-
n
o
i
s
i
n
g
t
ec
h
n
i
q
u
e
an
d
Pre
-
w
h
i
t
e
n
i
n
g
f
i
l
t
er,
"
A
p
p
l
i
e
d
A
co
u
s
t
i
c
s
,
v
o
l
.
1
2
3
,
p
p
.
9
3
-
1
0
6
,
2
0
1
7
.
[
22]
Z
.
Z
h
o
u
et
al
.
,
"
Imp
ro
v
emen
t
o
f
t
h
e
s
i
g
n
al
t
o
n
o
i
s
e
rat
i
o
o
f
L
i
d
ar
ech
o
s
i
g
n
a
l
b
as
e
d
o
n
w
av
e
l
e
t
d
e
-
n
o
i
s
i
n
g
t
ech
n
i
q
u
e,
"
O
p
t
i
c
s
a
n
d
La
s
er
s
i
n
E
n
g
i
n
eer
i
n
g
,
v
o
l
.
5
1
,
p
p
.
9
6
1
-
9
6
6
,
2
0
1
3
.
[
23]
D
.
L
.
D
o
n
o
h
o
,
"
D
e
-
n
o
i
s
i
n
g
b
y
s
o
f
t
-
t
h
re
s
h
o
l
d
i
n
g
,
"
IE
E
E
Tr
a
n
s
a
ct
i
o
n
s
o
n
In
f
o
r
m
a
t
i
o
n
Th
eo
r
y
,
vol.
4
1
,
n
o
.
3
,
p
p
.
6
1
3
-
6
2
7
,
1
9
9
5
.
[
24]
I.
M.
J
o
h
n
s
t
o
n
e
an
d
B.
W
.
Si
l
v
erman
,
"
W
av
el
e
t
t
h
re
s
h
o
l
d
es
t
i
mat
o
rs
fo
r
d
at
a
w
i
t
h
co
rrel
a
t
ed
n
o
i
s
e,
"
Jo
u
r
n
a
l
o
f
t
h
e
R
o
ya
l
S
t
a
t
i
s
t
i
ca
l
S
o
c
i
et
y
:
S
er
i
es
B
,
v
o
l
.
5
9
,
p
p
.
3
1
9
-
3
5
1
,
1
9
9
7
.
[
25]
R.
A
g
g
arw
a
l
,
J
.
K
.
Si
n
g
h
,
V
.
K
.
G
u
p
t
a,
S.
Rat
h
o
re,
M
.
T
i
w
ari
,
an
d
A
.
K
h
are
,
"
N
o
i
s
e
red
u
c
t
i
o
n
o
f
s
p
eec
h
s
i
g
n
al
u
s
i
n
g
w
a
v
el
e
t
t
ra
n
s
f
o
rm
w
i
t
h
m
o
d
i
fi
e
d
u
n
i
v
ers
a
l
t
h
re
s
h
o
l
d
,
"
In
t
er
n
a
t
i
o
n
a
l
Jo
u
r
n
a
l
o
f
Co
m
p
u
t
e
r
A
p
p
l
i
c
a
t
i
o
n
s
,
v
o
l
.
2
0
,
n
o
.
5
,
p
p
.
1
4
-
1
9
,
2
0
1
1
.
[
26]
C.
T
an
,
A
.
S
l
u
ze
k
,
G
.
S.
G
L
,
an
d
T
.
J
i
an
g
,
"
Ran
g
e
g
a
t
ed
i
mag
i
n
g
s
y
s
t
em
f
o
r
u
n
d
er
w
at
er
ro
b
o
t
i
c
v
eh
i
cl
e,
"
O
C
E
A
NS
2
0
0
6
-
A
s
i
a
P
a
ci
f
i
c
,
p
p
.
1
-
6
,
2
0
0
7
.
[
27]
C.
T
an
,
G
.
Seet
,
A
.
Sl
u
ze
k
,
an
d
D
.
H
e,
"
A
n
o
v
e
l
ap
p
l
i
cat
i
o
n
o
f
ra
n
g
e
-
g
at
e
d
u
n
d
er
w
at
er
l
as
er
i
mag
i
n
g
s
y
s
t
em
(U
L
IS)
i
n
n
ear
-
t
arg
e
t
t
u
r
b
i
d
med
i
u
m,
"
O
p
t
i
c
s
a
n
d
La
s
er
s
i
n
E
n
g
i
n
eer
i
n
g
,
v
o
l
.
4
3
,
n
o
.
9
,
p
p
.
9
9
5
-
1
0
0
9
,
2
0
0
5
.
[
28]
J
.
E
l
l
i
n
a
s
,
T
.
Man
d
a
d
el
i
s
,
A
.
T
zo
r
t
zi
s
,
an
d
L
.
A
s
l
a
n
o
g
l
o
u
,
"
Imag
e
d
e
-
n
o
i
s
i
n
g
u
s
i
n
g
w
a
v
el
e
t
s
,
"
TE
I
o
f
P
i
r
a
eu
s
A
p
p
l
i
ed
R
e
s
ea
r
ch
R
evi
ew
,
v
o
l
.
9
,
p
p
.
9
7
-
1
0
9
,
2
0
0
4
.
Evaluation Warning : The document was created with Spire.PDF for Python.